Compiler Explorer

Source code

source_filename = "matmul.c"
target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n32:64-S128"
target triple = "riscv64-unknown-unknown"

; Function Attrs: nofree norecurse nosync nounwind memory(argmem: readwrite)
define dso_local void @matmul(ptr noalias nocapture noundef readonly %0, ptr noalias nocapture noundef readonly %1, ptr noalias nocapture noundef %2) local_unnamed_addr #0 {
  br label %4

4:                                                ; preds = %3, %14
  %5 = phi i64 [ 0, %3 ], [ %15, %14 ]
  %6 = shl nuw nsw i64 %5, 6
  br label %8

7:                                                ; preds = %14
  ret void

8:                                                ; preds = %4, %17
  %9 = phi i64 [ 0, %4 ], [ %18, %17 ]
  %10 = or disjoint i64 %9, %6
  %11 = getelementptr inbounds float, ptr %2, i64 %10
  %12 = load float, ptr %11, align 4, !tbaa !6
  %13 = getelementptr float, ptr %1, i64 %9
  br label %20

14:                                               ; preds = %17
  %15 = add nuw nsw i64 %5, 1
  %16 = icmp eq i64 %15, 64
  br i1 %16, label %7, label %4, !llvm.loop !10

17:                                               ; preds = %20
  store float %29, ptr %11, align 4, !tbaa !6
  %18 = add nuw nsw i64 %9, 1
  %19 = icmp eq i64 %18, 64
  br i1 %19, label %14, label %8, !llvm.loop !12

20:                                               ; preds = %8, %20
  %21 = phi i64 [ 0, %8 ], [ %30, %20 ]
  %22 = phi float [ %12, %8 ], [ %29, %20 ]
  %23 = or disjoint i64 %21, %6
  %24 = getelementptr inbounds float, ptr %0, i64 %23
  %25 = load float, ptr %24, align 4, !tbaa !6
  %26 = shl nuw nsw i64 %21, 6
  %27 = getelementptr float, ptr %13, i64 %26
  %28 = load float, ptr %27, align 4, !tbaa !6
  %29 = tail call float @llvm.fmuladd.f32(float %25, float %28, float %22)
  %30 = add nuw nsw i64 %21, 1
  %31 = icmp eq i64 %30, 64
  br i1 %31, label %17, label %20, !llvm.loop !13
}

; Function Attrs: mustprogress nocallback nofree nosync nounwind speculatable willreturn memory(none)
declare float @llvm.fmuladd.f32(float, float, float) #1

attributes #0 = { nofree norecurse nosync nounwind memory(argmem: readwrite) "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="generic-rv64" "target-features"="+64bit,+a,+c,+d,+f,+m,+relax,+zicsr,+zifencei,-e,-experimental-zacas,-experimental-zcmop,-experimental-zfbfmin,-experimental-zicfilp,-experimental-zicfiss,-experimental-zimop,-experimental-ztso,-experimental-zvfbfmin,-experimental-zvfbfwma,-h,-smaia,-smepmp,-ssaia,-svinval,-svnapot,-svpbmt,-v,-xcvalu,-xcvbi,-xcvbitmanip,-xcvelw,-xcvmac,-xcvmem,-xcvsimd,-xsfvcp,-xsfvfnrclipxfqf,-xsfvfwmaccqqq,-xsfvqmaccdod,-xsfvqmaccqoq,-xtheadba,-xtheadbb,-xtheadbs,-xtheadcmo,-xtheadcondmov,-xtheadfmemidx,-xtheadmac,-xtheadmemidx,-xtheadmempair,-xtheadsync,-xtheadvdot,-xventanacondops,-za128rs,-za64rs,-zawrs,-zba,-zbb,-zbc,-zbkb,-zbkc,-zbkx,-zbs,-zca,-zcb,-zcd,-zce,-zcf,-zcmp,-zcmt,-zdinx,-zfa,-zfh,-zfhmin,-zfinx,-zhinx,-zhinxmin,-zic64b,-zicbom,-zicbop,-zicboz,-ziccamoa,-ziccif,-zicclsm,-ziccrse,-zicntr,-zicond,-zihintntl,-zihintpause,-zihpm,-zk,-zkn,-zknd,-zkne,-zknh,-zkr,-zks,-zksed,-zksh,-zkt,-zmmul,-zvbb,-zvbc,-zve32f,-zve32x,-zve64d,-zve64f,-zve64x,-zvfh,-zvfhmin,-zvkb,-zvkg,-zvkn,-zvknc,-zvkned,-zvkng,-zvknha,-zvknhb,-zvks,-zvksc,-zvksed,-zvksg,-zvksh,-zvkt,-zvl1024b,-zvl128b,-zvl16384b,-zvl2048b,-zvl256b,-zvl32768b,-zvl32b,-zvl4096b,-zvl512b,-zvl64b,-zvl65536b,-zvl8192b" }
attributes #1 = { mustprogress nocallback nofree nosync nounwind speculatable willreturn memory(none) }

!llvm.module.flags = !{!0, !1, !2, !4}
!llvm.ident = !{!5}

!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{i32 1, !"target-abi", !"lp64d"}
!2 = !{i32 6, !"riscv-isa", !3}
!3 = !{!"rv64i2p1_m2p0_a2p1_f2p2_d2p2_c2p0_zicsr2p0_zifencei2p0"}
!4 = !{i32 8, !"SmallDataLimit", i32 8}
!5 = !{!"Homebrew clang version 18.1.8"}
!6 = !{!7, !7, i64 0}
!7 = !{!"float", !8, i64 0}
!8 = !{!"omnipotent char", !9, i64 0}
!9 = !{!"Simple C/C++ TBAA"}
!10 = distinct !{!10, !11}
!11 = !{!"llvm.loop.mustprogress"}
!12 = distinct !{!12, !11}
!13 = distinct !{!13, !11}